Unstable readings in photoresistor matrix with Arduino, OK with Ohmmeter - what's wrong?

Question

I'm working on a photoresistor array with my arduino, and I'm having a little trouble getting stable readings.

I'm driving 7 rows one at a time with Arduino's digitalWrite(pin,HIGH) command and floating the other outputs with pinMode(pin,INPUT); Problem is, I got NO READING from my analogRead() voltage dividers unless the column is hooked up to my ohmmeter. Whats up with that?

Only the column connected to my ohmmeter outputs anything at all, all the other columns read zero. Whats in my ohmmeter that suddenly makes that sensor work?

So if I connect my Ohm-meter to the positive rail of P1 and to MEGA Analog 0 (A0) connection, suddenly the whole column gets readings! Otherwise, zeros everywhere.

void readArray() 
{ 
  for(int i=0; i < ROWS; i++)
  { 
    for(int k=0; k<ROWS; k++)
    { 
      pinMode(rowPins[k], INPUT); 
    } 

    pinMode(rowPins[i],OUTPUT); 
    digitalWrite(rowPins[i],HIGH); 
    delay(250); 

    for(int j=0; j < COLS; j++) 
    { 
      holdVals[i][j] = analogRead(j); 
      Serial.println(holdVals[i][j]); 
    } 

    digitalWrite(rowPins[i],LOW); 
  } 
}

Answer 1

The answer JustJeff gave you is probably correct for why you only get something when the ohmmeter is connected. I'd like to address a different issue. That issue is that your photoresistor array isn't going to work.

The problem is that you cannot "turn on" a single row at a time. For example, lets say that you turn on the middle row. The top and bottom row lines are tri-stated from the uC. Current will go from the middle row line, through P5 to the left most column, through P1 to the top row line, through p2 to the next column line, and mess up your measurements for that column. Now imagine lots of "false paths" and you can probably work out that everything will get hopelessly messed up.

One solution to this is to put a schottky diode in series with every photoresistor (with the diode "pointed" to the column lines). You read a row just like you do now, although you might try driving the "unused rows" with a low signal instead of tri-stating them. I'm not sure if you'll have some wacky leakage current issues if you don't.

Even without the diodes, however, I would have expected you to read at least some voltage. Your low-side resistors (R in your schematics) might be too low of a value. Do like JustJeff said and measure the voltage on a column line. If it's close to zero then you have a hardware problem. If it's not close to zero then you have a software problem.

P.S., I didn't look at your C code, since I'm not familiar with programming on the Arduino.

Answer 2

When set to measure ohms, your meter actually applies a little voltage to the item under test. Ordinarily this results in a current that the meter can then use to infer the resistance being measured. It just so happens that it's also providing a voltage drop that your ADC input can observe as well.

What you should do is set the meter for voltage readings, and see if your digitalWrite() is having the intended effect. Sure seems like it isn't. I forget (off the top of my head), but you may have to actually configure the pin for output in order for digitalWrite() to work. Maybe not.

Answer 3

A couple of things:

1. You are not supposed to measure resistance using an ohm-meter with power applied to your circuit as a general rule.

2. On the software side of things, pinMode set to INPUT can be a little bit misleading. AVR pins can actually have three modes: output, high impedance input, and internally weakly pulled up input. When you specify pinMode INPUT you are in one of the latter two cases, because all it does is affect the DDRx register. To ensure it is in fact a high impedance input, you should also digitalWrite LOW the same pin, which affects the PORTx register.

In effect what is happening is when you set your outputs HIGH for the first time, the next time you turn them to inputs, the internal passive pullups are enabled. So, in summary, instead of pinMode(rowPins[k], INPUT); use two statements:

pinMode(rowPins[k], INPUT);
digitalWrite(rowPins[k], LOW);

Answer 4

Lots of good suggestions but I would suggest a few low-hanging fruit:

1) bad connections / loose wires so the rows aren't energized;

2) wrong pins connected vs. the code;

I would try to rule them out before digging deeper.

decoding the resistor values based on adc readings can be a challenge, given how the matrix is wired.